Kaypro 4 '83 - Turns on but doesn't boot. Pop sound and smoke.

[Pyroteq]

A customer of mine gave me a Kaypro 4 '83 he had lying around. He told me that it still works, although who knows when it was turned on last.

I tried turning it on and was presented with what I can only describe as the Matrix - random characters rapidly changing filling the entire screen. I believe the power LED as well as both floppy indicator LED's were on. A few seconds later I heard a pop and saw smoke coming from the chassis, so I quickly turned it off.

Now I'm good at fixing modern computers but this thing is from before I was born, so I don't know where to start with this one. I'd like to get it running again, no idea what I'll do with it but I've always liked that luggable computer look and the monochrome CRT is pretty sweet.

I opened it up to inspect for any damage, but couldn't see anything obvious. Doing some reading it seems recommended to remove the RIFA capacitors. I removed 4 (3 small and one larger) off the Boschert power supply although from a visual inspection it wasn't obvious that any of those had been the cause of the pop and smoke but figured it was best to be safe.



After removing those caps I tried turning it on again but this time I'm just presented with a screen filled with

'N'N'N'N'N'N'N'N'N'N

repeating over the entire screen.



I'm not sure if that's worse than before or an improvement. As before the power LED's and the floppy disk indicator lights were on.

I thought I remembered smoke coming from the right side of the chassis when it had made the pop sound, so I thought the next culprit may be some sort of short on the floppy drives, blown cap, etc, so I disconnected them entirely from the system and attempting to boot it again but got the same repeating 'N pattern as before.

So not sure where to go to next. Could it be a power supply issue? I could try replacing with the MEAN WELL RPT 75B I have seen recommended but I have no idea if a power supply issue would cause that sort of behaviour on boot.

There was one area of the power supply that looked a bit dodgy to me as well, I wasn't sure if it's melted solder thats caused a short of if that was scraped away during manufacture for some reason as I don't often come across PCB's this old that look hand made.



I've attached relevant pictures.

Thank you for any advice you can provide.

 

[ajacocks]

Welcome to the forum @Pyroteq .

I am far from expert in PSU repair. However. I would start by checking voltages from the power supply output.

The fact that you are getting good high voltage (crt works) and that you are getting characters on the screens says to me that things are close to working.

- Alex

 

[rlauzon]

The Kaypros have RIFA caps and they usually fail after this long a time.

 

[daver2]

I assume the keyboard is connected to the machine?

The most likely fault is now the keyboard.

Can you disconnect the keyboard to see what happens?

Dave

 

[norwestrzh]

Pyroteq:

Interesting ID, BTW! <grin> From your description (pop and smoke), I'd guess a capacitor blew up somewhere. Per the other replies, detach as much as you can -- keyboard, even floppies. See what gives then. A failed cap. might not even be visible with casual inspection. Got a voltmeter? Check for shorts at the caps. and from the +5 and +12v rails to ground. Did you see the discussion elsewhere here about bad caps in Kaypro floppy drives?

Roger

 

[Pyroteq]

This isn't my footage, but I found a video that demonstrates similar to what I saw the first time I had powered it on.

I'm assuming something I've done has made it go from that to the repeating 'N pattern. I've pull it apart and reconnected cables a few times so I guess if there's some corrosion on the connectors I guess maybe that could explain it. I read somewhere it was fairly common for those connections to become very flaky after so many years and it was even recommended to just solder the wires onto the power supply.

I assume the keyboard is connected to the machine?

No, the jack on the back of the keyboard is damaged and rattling around inside the keyboard so I'll need to order a replacement part for that. Wanted to check the machine even functions before I place an order.

detach as much as you can -- keyboard, even floppies. See what gives then

Yup, same result when floppy drives and keyboard are disconnected.

The Kaypros have RIFA caps and they usually fail after this long a time.

Removed all of them already off the power supply. Not sure if that's what the pop was from though as I could see cracks on all of them, but no obvious scorch marks.

Check for shorts at the caps. and from the +5 and +12v rails to ground.

I just used my multimeter to check the voltages on the motherboard power pins. I was getting about 12.20v on the +12v, 11.98 on the -12v and 5.05 on the 5v, so they all seem to be within what I'd expect. I was testing the voltages while powered on.


If I hit the reset button while powered on what should I expect to happen? Should the screen go blank for a few seconds? As far as I could tell it wasn't doing anything I could notice, the repeating 'N pattern stayed on the screen.

 

[Pyroteq]

I read elsewhere the reset button can become an issue if it's grounded so I attempted booting the computer with the reset button disconnected from the motherboard but that made no difference.

Unfortunately we have a casualty.

I noticed the memory chips seemed pretty corroded and wanted to see if I could clean them up but the leg snapped off this Mostek MK4564N chip. =(

I tried to be careful pulling it out, but it was stuck and bent the leg as it came out and due to how brittle it was it just snapped clean off.



For some strange reason this entire column of chips from U20 to U27 has some pretty serious rust on the legs while the rest of them look fine. Perhaps the metal on these legs are different to the metal on all the other chips?



Compare the condition of those to the rest which seem to be in pretty good condition.



Google suggests these are 64K memory chips?

It seems there's 2 different numbers on them, the MOSTEK 8304 as well as the 8248, am I correct in assuming these are simply date stamps or something and all these are all the same MK5464N-25 chip?

Maybe worth replacing the lot of them if it's affordable due to the heavy amount of corrosion on them. I have a feeling I'm going to end up with more broken chips if I try pulling them all out and cleaning them.

 

[daver2]

Yes, different metal or finish. Been there got the T-shirt!

The 4-digit number is the date code of manufacture.

Yes, 64K bit memory chip. Notice the -25 on the end. That is the speed rating. Don't use anything slower than a -25 or it may not work.

Check the data sheet for these devices and the ones you are thinking of buying and look for differences. Post anything you don't understand.

Also double check for 128 or 256 cycle refresh. You need to replace with the same type to be safe.

Dave

 

[Pyroteq]

Thanks for that info. I think the memory is probably the best place to start at this point. Does the motherboard require every memory slot populated or can it run with fewer sticks? Perhaps I could try salvaging as many as I can and cleaning them just to see if I can get it to boot and then purchase the remaining memory chips.

Or is this a matter of just buying a whole set as it won't function without all of them?

As for the speed, this is a matter of lower is better right as it's measured in ms/ns, so a 20 or 15 could also work if they all match?

 

[Hugo Holden]

The Rifa caps failed because as you can see, they have multiple cracks in their plastic cases, this lets in moisture and the paper insulation swells up (they are metalized paper types) Then they become leaky electrically, draw current and smoke. Just remove them, there is no need to replace them. In this day and age any SMPS wall wart supply will inject x10 the amount of RF hash onto the power line than your computer supply, and fortunately your computer supply can ignore the current levels of radio frequency interference from other modern PSU's contaminating the line power with interference, even without the Rifa caps present. You could put replacement Rifa caps back in, but in 30 to 40 years some technician will be wondering why they are smoking.

Mostly, SMPS supplies cause annoying interference on the AM (MW and SW) radio bands (so do people's home Wifi systems now), but manufacturers now don't care. It has become apparent that the number of people listening to AM radio are minimal and they are aging/dying out rapidly. In fact I don't know anybody in my entire area locally that listens to AM radio except me. So the manufacturers know they will get minimal complaints about their equipment interfering with AM radio reception. Also the younger generation barely know what a tuning knob is for on an AM radio and feel it is a useless device which cannot navigate the internet, where practically all radio stations are anyway these days.

Your power supplies are working now which is good.

Many IC types, specifically TI ones used steel pins. If you check you will find they are magnetic. But in TI's case, they coated the pins with Silver. (many other IC's use Tin plating which is better at sacrificing electrons to the steel and many other's had a TIN-LEAD coat for easy soldering and were more rust & corrosion resistant).

In any case, once the Silver on the surface fails, rust sets in and then it is not uncommon for the legs to rust and break off when the IC's are removed.

 

[cj7hawk]

Yes, as noted, the RIFA capacitors are a problem. Replace or remove. They are only filters. The PSU will likely work. You can also get smoke from the floppy drives - check the voltage rails to make sure there are no shorts there.

Check voltages. You should have -12, +12 and +5v from the PSU.

You have an early motherboard. It's not the 64K chips that are generating the video - They might be faulty - but if the display doesn't change and isn't random on later power up, then it's probably not related to the CPU not changing.

Your video memory is 4 x 2114 static ram chips ( 4 x 2k each ).



If the CPU isn't starting, it won't clear the screen. The video display is generated in hardware in your board, so will produce whatever is in the Video RAM. it might be the main ram, but could be just about anything on the main board causing that display.

Check that the BOOT ROM is being accessed and is supplying data to the CPU. @durgadas311 is the Kaypro expert and has some test ROMs he has written - it would be worth asking him about the easy way to test the Kaypro if it's not starting -

David.

Also, what kinds of tools do you haave?

 

[djg]

Does the motherboard require every memory slot populated or can it run with fewer sticks?

The chips are 64k by 1 bit so you need all 8 to make a byte.

As for the speed, this is a matter of lower is better right as it's measured in ms/ns, so a 20 or 15 could also work if they all match?

Correct. There is always the chance with poor design that faster chips can cause problems but don't think its that common.

 

[cj7hawk]

Google suggests these are 64K memory chips?

I think you can use any 4164 type RAM chip, but you can get those too relatively cheaply. Eg: https://vi.aliexpress.com/item/1005005429915900.html

Speed should not be a problem. Faster ( lower number) is usually better.

It's always good to have the same type, but in truth you can usually mix them up with different types as long as they are all fast enough.

My rams are also a bit "Rusty" - But they work OK. It only seems to affect the RAM chips in my Kaypro.

 

[Pyroteq]

Well I took out all the memory chips. Seems using a flathead screwdriver and prying from beneath is much safer than trying to wiggle them out with my fingers. Lesson learned... Managed to get the rest out without damage and gave them a good clean.

Before:



Side by side:



After:




I sat the legs in some vinegar for a few minutes, then used a nail file to remove as much rust as I could, then used 99% IPA to wash off the residue. Came out looking much better so perhaps I can continue to use these after all if I can solder a new leg onto the IC I broke.

Unfortunately after all that work I didn't get any different result. There's a chance the memory might have been messed up before, but I wasn't getting far enough in the boot process to notice it yet so perhaps this is just a preemptive step for later. Besides, a good idea to get that rust out of the sockets. The sockets still seemed in good shape considering the gross gunk that was sitting in them.

Also, what kinds of tools do you haave?

Screwdrivers, etc.
Soldering iron
Multimeter
Raspberry Pi 3B (Not sure if that's useful in this situation, I recently used it to serial into a security camera I'm attempting to fix via UART pads)
Arduino starter kit with Arduino Uno R3 (Apparently these can be used to write bios with? I've actually never used this, I bought it for my kids to play with so I'm a complete newb to it but I can follow a guide)

I don't have an oscilloscope or any other retro computers I can salvage parts from. I think the next oldest working motherboard I have lying around is a Pentium 3 or something.

Anyway, I read something before about being careful about IC stickers? I didn't pay much attention to it, but it jolted my mind so I did a bit of further reading and came across this quote:

This is a picture of the ROM. When I first opened the Kaypro, there was nothing on the top of the IC. This is dangerous, as exposure to sunlight could wipe the chip!

Apparently this is news to me! Never heard of chips losing their data when exposed to light before. What kind of black magic do these old machines run on?! They're vampires that die when exposed to light?

Well just so happens there's an IC with a sticker missing up the top. At some stage I had also taken the machine outside and used my electronic air duster to blast out all the dust, webs and other crap. So it definitely got exposed to sunlight... Not sure if the sticker was there when I first opened up the machine or I blasted it off with air when cleaning it.



Looking up a schematic that's the character generator EPROM, is that correct?

I was told the machine worked before I turned it on (that could have been a decade ago), however I did open the chassis first to ensure the 220v jumper was selected on the power supply as the back of the chassis had a 110v label and wasn't sure if the previous owner used a step down transformer so it may have even been wiped when I did this if it's super sensitive to light.

I have no idea if about 2 minutes outside would screw it up, but perhaps this is the issue?

If so, would it be possible to verify and/or reprogram this chip using an arduino/raspberry pi and a breadboard?

Can any of this save me?

 

[daver2]

You would have to stand out in the bright sunlight all day every day for many years! So a bit of light on an exposed EPROM isn't going to cause problems.

The issue is that EPROMs are erased with Ultra Violet (UV) light - but not sunlight, we are talking high energy UV.

EPROMs can, however, degrade over time via a process called "bit rot". EPROMs (and ROMs) were not designed to last this long...

Dave

 

[cj7hawk]

Unfortunately after all that work I didn't get any different result. There's a chance the memory might have been messed up before, but I wasn't getting far enough in the boot process to notice it yet so perhaps this is just a preemptive step for later. Besides, a good idea to get that rust out of the sockets. The sockets still seemed in good shape considering the gross gunk that was sitting in them.

They are old chips. Cheap enough to order replacements. You will need some good ones to get things working.

Screwdrivers, etc.
Soldering iron
Multimeter
Raspberry Pi 3B (Not sure if that's useful in this situation, I recently used it to serial into a security camera I'm attempting to fix via UART pads)
Arduino starter kit with Arduino Uno R3 (Apparently these can be used to write bios with? I've actually never used this, I bought it for my kids to play with so I'm a complete newb to it but I can follow a guide)

You probably also want to get a DRAM tester, since any DRAM you now get is going to be suspected faulty until proven otherwise. They are cheap on Ebay.

Also an oscilloscope is going to be nearly a must... You might get away with a logic probe, but on an older machine with no known working state, an oscilloscope is nearly a must. Look for something at least around 100 Msps - a cheap one will do, though if you're going to end up fixing stuff, most likely you will fix other stuff in the future, so consider whether buying a slightly better one is worthwhile. At the lower end, some multimeters come with a built in oscilloscope function now which would be suitable and cost around $USD 60. Make sure to look for a Dual Trace if possible. It helps a lot.

You will also need an eprom programmer and some blank eproms, but not for the reason you're assuming. ( You don't *require* them, but it will make your task easier ). This will be the most expensive part you need to purchase to repair old computers, unless you buy a very good oscilloscope.

I don't have an oscilloscope or any other retro computers I can salvage parts from. I think the next oldest working motherboard I have lying around is a Pentium 3 or something.

Anyway, I read something before about being careful about IC stickers? I didn't pay much attention to it, but it jolted my mind so I did a bit of further reading and came across this quote:

As was mentioned, this isn't really a problem - the only issue removing the labels is that they are a part of the original machine, hence desireable. Sunlight can erase them, but only if you're in space. The Atmosphere attenuates the light wavelength they are sensitive to until it's practically non-existant at the surface of the earth.... Maybe if you lived on a mountain it might be a problem to leave them without labels for a few years in direct sunlight. I've tried to erase these chips with sunlight many times, not even a bit got damaged. But it's not impossible, so they say keep labels on things, though usually we put labels on them so you'd know they were already programmed as the assumption is unlabeled chips are blank.

Looking up a schematic that's the character generator EPROM, is that correct?

Yes, and it appears to be working OK. You can assume it's fine. Or at least, it's program is fine. It could potentially have an electrical fault but that is unlikely.

I was told the machine worked before I turned it on (that could have been a decade ago), however I did open the chassis first to ensure the 220v jumper was selected on the power supply as the back of the chassis had a 110v label and wasn't sure if the previous owner used a step down transformer so it may have even been wiped when I did this if it's super sensitive to light.

It might have, but who knows when that was? No sense worrying about it. Sometime people think "working" means "light came on"... It doesn't mean operates correctly anymore.

I have no idea if about 2 minutes outside would screw it up, but perhaps this is the issue?

Not unless it was raining for those two minutes.

OK, so you need to confirm z80 function first... Track down a user named @durgadas311 and ask him very nicely if he can send you a copy of a test ROM for a Kaypro. It sends diagnostic information out of the serial port and can be used to test chips electronically and directly. As I mentioned, he's a kaypro expert.

Your first challenge is to make sure enough of the computer is working so that the z80 has a valid clock, is generating addresses and read requests, and these are decoding correctly and the buffers are working between the ROM and the CPU that it starts to execute. If this much isn't working, then you won't get anywhere.
Also, download a full copy of the Kaypro schematics. I think your board is a Kaypro II, but again, I'm not an expert on it, but you appear to have the board before they put in the CRTC chip. Kaypros came with mixed boards a lot apparently.

Try this link -

Kaypro / Information & Downloads // retrocmp / retro computing

Kaypro Information

retrocmp.de

David,

 

[cj7hawk]

Oh, these Eproms are a bit funny about voltages - Try to find a programmer that does 21v and 25v. You can also use larger eproms, but you will need to hand wire up an adapter to fit them at times.

 

[cj7hawk]

Sorry, one other thing to add - if you're not familiar with an oscilloscope, you can also get a cheap 8 bit logic analyzer. Not as useful as an oscilloscope in some ways, but better in others since you can take your time to view the signals.

These are dirt cheap, and very useful for z80 type work, but you're going to need to buy leads and connectors to go onto chips. But these will let you check multiple lines at once even if it takes a long time to set them up.

 

[Pyroteq]

OK, so you need to confirm z80 function first... Track down a user named @durgadas311 and ask him very nicely if he can send you a copy of a test ROM for a Kaypro. It sends diagnostic information out of the serial port and can be used to test chips electronically and directly. As I mentioned, he's a kaypro expert.

I've found their website which has some good documentation.

http://sebhc.durgadas.com/kaypro/monitor.pdf

Includes a memory test ROM too.

I have null modems, serial cables, etc, so that seems like a cool tool to check out if I can get my hand on some blank EEPROMS - I'd prefer to use those because I won't need to worry about higher voltages to write them from what I've been reading.

As an update I don't know if I've made things better or worse...



I pulled out the boot rom down the bottom of the board, as well as the Z80 CPU and had a check for corrosion. After putting them back in place I am now getting the Matrix effect again, so it seems that's done something.

Later on I'll attempt to carefully pull every single one out and reseat them. Kinda wanted to avoid that since they're so old and fragile, but if moving 2 chips has made a difference there's a good chance moving some more around might help.

Some of the characters stay static on the screen while the rest are flashing.

As one of the memory chips is damaged I also swapped it to another location just to see if that had an effect on what appeared on screen, but I don't believe it did.

 

[daver2]

I think your difference is a retrograde step though ...

Most of these vintage computers have dynamic RAM for the main computer and static RAM for the display. The display logic is mostly hardwired.

As a result, if you turn the computer on without the CPU running, you generally get a semi random display.

It requires the CPU to be running, and executing the correct instructions stored in the ROMs, to clear the display.

If you were getting a load of 'N characters on the screen, it probably indicates to me that they got there via the CPU.

However, if you are now getting a screenfull of semi random characters, it probably indicates that the CPU is not running properly at all, whereas before it was doing something - just not perhaps the correct thing...

Dave